Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'

import pickle as pkl

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x1e577925400>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x1e577a480f0>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.5.0
Default GPU Device: /device:GPU:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests
import tensorflow as tf

# Rank means t
def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    input_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name="input_real")
    input_z = tf.placeholder(tf.float32, (None, z_dim), name="input_z")
    learning_rate = tf.placeholder(tf.float32, [], name="learning_rate")
    
    return input_real, input_z, learning_rate

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    
    # Drop out rate
    drop_rate = 0.1
    # Alpha rate for leaky relu
    alpha = 0.2
    kernel_size = 3
    
    n_units = 128
    size_mult = 64
    x = images    
    
    with tf.variable_scope("discriminator", reuse=reuse):
        
        x1 = tf.layers.conv2d(x, size_mult, kernel_size, strides=2, padding='same',
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        dropout1 = tf.layers.dropout(x1, rate=drop_rate)
        relu1 = tf.maximum(alpha * dropout1, dropout1)
        
        x2 = tf.layers.conv2d(relu1, size_mult * 2, kernel_size, strides=2, padding='same',
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        bn2 = tf.layers.batch_normalization(x2, training=True)
        dropout2 = tf.layers.dropout(bn2, rate=drop_rate)
        relu2 = tf.maximum(alpha * dropout2, dropout2)
      
        x3 = tf.layers.conv2d(relu2, size_mult * 4, kernel_size, strides=2, padding='same',
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)

        # Flattening conv layer  
        features = tf.reshape(relu3, [-1, 4 * 4 * 256])
        
        # Set class_logits to be the inputs to a softmax distribution over the different classes
        class_logits = tf.layers.dense(features, 1, activation=None)
        
        # Sigmoid and not softmax. Sigmoid because only 2 classes.
        out = tf.nn.sigmoid(class_logits)
        
    return out, class_logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    
    reuse = not is_train
    alpha = 0.2
    n_units = 128
    kernel = 5
    stride = 2
    
    with tf.variable_scope("generator", reuse=reuse):
        
        # fully connected layer
        fc = tf.layers.dense(z, 2 * 2 * n_units * 4, activation=None)
        x1 = tf.reshape(fc, [-1, 2, 2, n_units * 4])   
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        
        x2 = tf.layers.conv2d_transpose(x1, n_units * 2, kernel, strides = stride, padding="valid",
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)

        x3 = tf.layers.conv2d_transpose(x2, n_units, kernel, stride, padding="same",
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, kernel, strides=stride, padding="same",
                             kernel_initializer=tf.contrib.layers.xavier_initializer())
        
        out = tf.tanh(logits)

        
    return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    g_model = generator(input_z, out_channel_dim)
    
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)
    
    # discriminator loss for the reals
    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_real, 
            labels=tf.ones_like(d_model_real)
        )
    )
    
    # Discriminator loss for the fakes
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_fake,
            labels=tf.zeros_like(d_model_fake)
        )
    )
    
    # Summing the discriminator loss for reals and fake
    total_d_loss = d_loss_real + d_loss_fake
    
    # Generator loss
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_fake, 
            labels=tf.ones_like(d_model_fake)
        )
    )
    
    return total_d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
        
    # NN vars
    disc_vars = tf.trainable_variables("discriminator")
    gen_vars = tf.trainable_variables("generator")

    # Optimize 
    disc_trained = tf.train.AdamOptimizer(learning_rate, beta1).minimize(d_loss, var_list = disc_vars)
        
    updates = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    g_updates = [opt for opt in updates if opt.name.startswith("generator")]
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=gen_vars)
    
    d_train_opt = disc_trained
    
    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

import time

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    #saver = tf.train.Saver()
    sample_z = np.random.normal(-0.8, 1, size=(batch_size, z_dim))

    samples, train_accuracies, test_accuracies = [], [], []
    steps = 0    
    
    print_every = 50
    show_every = 250
    images_to_show = 25
    
    print(data_shape)
    width, height, channels = data_shape[1:]
    print("width: {}, height: {}, channels: {}".format(width, height, channels))
    print("Channel mode: {}".format(data_image_mode))
    
    
    # Getting the variables required for training 
    input_real, input_z, some_learning_rate = model_inputs(width, height, channels, z_dim)
    
    # Should use the channel mode to calculate the channels but using the number of channels works too
    d_loss, g_loss = model_loss(input_real, input_z, channels)
    
    # Optimization
    d_train_opt, g_train_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for e in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                            
                batch_images *= 2.0
                    
                steps += 1
                
                # Sample random noise for G
                batch_z = np.random.uniform(-0.8, 1, size=(batch_size, z_dim))
                
                # Run the optimizer for discriminator
                _ = sess.run(
                    d_train_opt, 
                    feed_dict={
                        input_real: batch_images, 
                        input_z: batch_z,
                        some_learning_rate:learning_rate
                    }
                )
                
                # Run the optimizer for the generator
                _ = sess.run(
                    g_train_opt,
                    feed_dict={
                        input_z: batch_z, 
                        input_real: batch_images,
                        some_learning_rate:learning_rate
                    }
                )
                
                # Print out the losses
                if steps % print_every == 0:
                    train_loss_d = d_loss.eval({
                        input_real: batch_images,
                        input_z: batch_z
                    })
                    train_loss_g = g_loss.eval({
                        input_z: batch_z, 
                        input_real: batch_images
                    })
                    print("Epoch {}/{}...".format(e+1, epochs),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                
                # Show the generator output
                if steps % show_every == 0:
                    show_generator_output(
                        sess,
                        images_to_show,
                        input_z, 
                        channels,
                        data_image_mode
                    )
        #saver.save(sess, './checkpoints/generator.ckpt')

    with open('samples.pkl', 'wb') as f:
        pkl.dump(samples, f)

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 16
z_dim = 112
learning_rate = 0.00045
beta1 = 0.3

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 5

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

(60000, 28, 28, 1)
width: 28, height: 28, channels: 1
Channel mode: L
Epoch 1/5... Discriminator Loss: 1.4317... Generator Loss: 0.3957
Epoch 1/5... Discriminator Loss: 0.2764... Generator Loss: 1.7129
Epoch 1/5... Discriminator Loss: 0.5306... Generator Loss: 1.9253
Epoch 1/5... Discriminator Loss: 0.7985... Generator Loss: 1.0396
Epoch 1/5... Discriminator Loss: 1.4452... Generator Loss: 0.4065

Epoch 1/5... Discriminator Loss: 0.6322... Generator Loss: 1.2927
Epoch 1/5... Discriminator Loss: 1.7811... Generator Loss: 0.3201
Epoch 1/5... Discriminator Loss: 0.4734... Generator Loss: 1.6226
Epoch 1/5... Discriminator Loss: 1.1455... Generator Loss: 0.5395
Epoch 1/5... Discriminator Loss: 0.9013... Generator Loss: 1.3247

Epoch 1/5... Discriminator Loss: 0.6647... Generator Loss: 1.2202
Epoch 1/5... Discriminator Loss: 1.1843... Generator Loss: 0.5920
Epoch 1/5... Discriminator Loss: 1.0210... Generator Loss: 1.1831
Epoch 1/5... Discriminator Loss: 1.4676... Generator Loss: 0.3669
Epoch 1/5... Discriminator Loss: 1.0296... Generator Loss: 0.9444

Epoch 1/5... Discriminator Loss: 1.3309... Generator Loss: 1.7406
Epoch 1/5... Discriminator Loss: 1.4865... Generator Loss: 0.3780
Epoch 1/5... Discriminator Loss: 1.7250... Generator Loss: 0.3920
Epoch 1/5... Discriminator Loss: 0.8529... Generator Loss: 1.0204
Epoch 1/5... Discriminator Loss: 1.4659... Generator Loss: 0.3366

Epoch 1/5... Discriminator Loss: 1.1752... Generator Loss: 0.6024
Epoch 1/5... Discriminator Loss: 1.2981... Generator Loss: 0.4080
Epoch 1/5... Discriminator Loss: 1.4226... Generator Loss: 0.3994
Epoch 1/5... Discriminator Loss: 1.1454... Generator Loss: 1.5889
Epoch 1/5... Discriminator Loss: 1.1051... Generator Loss: 1.3249

Epoch 1/5... Discriminator Loss: 1.9756... Generator Loss: 0.1886
Epoch 1/5... Discriminator Loss: 1.0465... Generator Loss: 0.7482
Epoch 1/5... Discriminator Loss: 1.0746... Generator Loss: 0.7871
Epoch 1/5... Discriminator Loss: 1.0588... Generator Loss: 0.7369
Epoch 1/5... Discriminator Loss: 1.1824... Generator Loss: 0.4932

Epoch 1/5... Discriminator Loss: 1.3362... Generator Loss: 0.4575
Epoch 1/5... Discriminator Loss: 0.6888... Generator Loss: 1.1685
Epoch 1/5... Discriminator Loss: 0.9174... Generator Loss: 0.7005
Epoch 1/5... Discriminator Loss: 1.2807... Generator Loss: 0.4225
Epoch 1/5... Discriminator Loss: 0.4168... Generator Loss: 1.3617

Epoch 1/5... Discriminator Loss: 0.9613... Generator Loss: 0.5841
Epoch 1/5... Discriminator Loss: 4.0328... Generator Loss: 0.0399
Epoch 1/5... Discriminator Loss: 0.6171... Generator Loss: 1.0884
Epoch 1/5... Discriminator Loss: 0.7198... Generator Loss: 0.8300
Epoch 1/5... Discriminator Loss: 0.6681... Generator Loss: 0.9975

Epoch 1/5... Discriminator Loss: 0.7201... Generator Loss: 1.5430
Epoch 1/5... Discriminator Loss: 0.5411... Generator Loss: 1.1766
Epoch 1/5... Discriminator Loss: 0.3431... Generator Loss: 1.5610
Epoch 1/5... Discriminator Loss: 0.7858... Generator Loss: 1.1215
Epoch 1/5... Discriminator Loss: 0.6158... Generator Loss: 1.0930

Epoch 1/5... Discriminator Loss: 2.3901... Generator Loss: 0.1603
Epoch 1/5... Discriminator Loss: 0.5727... Generator Loss: 1.0841
Epoch 1/5... Discriminator Loss: 0.4817... Generator Loss: 1.2148
Epoch 1/5... Discriminator Loss: 0.3333... Generator Loss: 1.6634
Epoch 1/5... Discriminator Loss: 1.7690... Generator Loss: 6.3838

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 16
z_dim = 112
learning_rate = 0.00045
beta1 = 0.3


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

(202599, 28, 28, 3)
width: 28, height: 28, channels: 3
Channel mode: RGB
Epoch 1/2... Discriminator Loss: 0.6263... Generator Loss: 0.9208
Epoch 1/2... Discriminator Loss: 0.6502... Generator Loss: 3.9704
Epoch 1/2... Discriminator Loss: 0.3629... Generator Loss: 3.1411
Epoch 1/2... Discriminator Loss: 1.1793... Generator Loss: 0.6624
Epoch 1/2... Discriminator Loss: 1.4490... Generator Loss: 0.5647

Epoch 1/2... Discriminator Loss: 1.5301... Generator Loss: 0.5149
Epoch 1/2... Discriminator Loss: 0.9182... Generator Loss: 1.0288
Epoch 1/2... Discriminator Loss: 1.0998... Generator Loss: 0.8182
Epoch 1/2... Discriminator Loss: 2.3403... Generator Loss: 0.2406
Epoch 1/2... Discriminator Loss: 1.2062... Generator Loss: 0.4959

Epoch 1/2... Discriminator Loss: 0.7884... Generator Loss: 1.1840
Epoch 1/2... Discriminator Loss: 1.0022... Generator Loss: 1.4590
Epoch 1/2... Discriminator Loss: 0.8068... Generator Loss: 1.5511
Epoch 1/2... Discriminator Loss: 1.5717... Generator Loss: 0.3573
Epoch 1/2... Discriminator Loss: 1.8697... Generator Loss: 0.2799

Epoch 1/2... Discriminator Loss: 0.7793... Generator Loss: 1.1788
Epoch 1/2... Discriminator Loss: 1.3698... Generator Loss: 0.5184
Epoch 1/2... Discriminator Loss: 0.6154... Generator Loss: 1.4457
Epoch 1/2... Discriminator Loss: 0.9647... Generator Loss: 0.9074
Epoch 1/2... Discriminator Loss: 1.0844... Generator Loss: 1.0791

Epoch 1/2... Discriminator Loss: 1.8534... Generator Loss: 0.2363
Epoch 1/2... Discriminator Loss: 1.0470... Generator Loss: 1.2995
Epoch 1/2... Discriminator Loss: 0.6708... Generator Loss: 1.3346
Epoch 1/2... Discriminator Loss: 0.5618... Generator Loss: 1.2223
Epoch 1/2... Discriminator Loss: 1.1241... Generator Loss: 0.8906

Epoch 1/2... Discriminator Loss: 1.0024... Generator Loss: 0.8012
Epoch 1/2... Discriminator Loss: 1.4363... Generator Loss: 0.6561
Epoch 1/2... Discriminator Loss: 0.6675... Generator Loss: 1.8961
Epoch 1/2... Discriminator Loss: 1.4127... Generator Loss: 0.4051
Epoch 1/2... Discriminator Loss: 1.0638... Generator Loss: 0.7699

Epoch 1/2... Discriminator Loss: 0.9174... Generator Loss: 1.2867
Epoch 1/2... Discriminator Loss: 1.4276... Generator Loss: 2.5133
Epoch 1/2... Discriminator Loss: 0.9901... Generator Loss: 0.9015
Epoch 1/2... Discriminator Loss: 0.6294... Generator Loss: 1.5032
Epoch 1/2... Discriminator Loss: 1.2990... Generator Loss: 0.5593

Epoch 1/2... Discriminator Loss: 0.9553... Generator Loss: 0.9541
Epoch 1/2... Discriminator Loss: 0.9541... Generator Loss: 1.1716
Epoch 1/2... Discriminator Loss: 0.7316... Generator Loss: 1.5085
Epoch 1/2... Discriminator Loss: 0.7699... Generator Loss: 1.2953
Epoch 1/2... Discriminator Loss: 0.9723... Generator Loss: 1.1995

Epoch 1/2... Discriminator Loss: 1.5505... Generator Loss: 0.4276
Epoch 1/2... Discriminator Loss: 1.4784... Generator Loss: 0.4449
Epoch 1/2... Discriminator Loss: 1.2842... Generator Loss: 0.4963
Epoch 1/2... Discriminator Loss: 0.7434... Generator Loss: 1.1106
Epoch 1/2... Discriminator Loss: 1.3095... Generator Loss: 0.6001

Epoch 1/2... Discriminator Loss: 1.4993... Generator Loss: 0.3948
Epoch 1/2... Discriminator Loss: 1.5139... Generator Loss: 1.7546
Epoch 1/2... Discriminator Loss: 1.4821... Generator Loss: 0.5510
Epoch 1/2... Discriminator Loss: 1.2481... Generator Loss: 0.5266
Epoch 1/2... Discriminator Loss: 0.7151... Generator Loss: 1.2799

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.